The loss of striatal dopamine (DA) in Parkinson's disease (PD) models triggers a cell-type-specific reduction in the density of dendritic spines in D2 receptor-expressing striatopallidal medium spiny neurons (D2 MSNs). How the intrinsic properties of MSN dendrites, where the vast majority of DA receptors are found, contribute to this adaptation is not clear. To address this question, two-photon laser scanning microscopy (2PLSM) was performed in patch-clamped mouse MSNs identified in striatal slices by expression of green fluorescent protein (eGFP) controlled by DA receptor promoters. These studies revealed that single backpropagating action potentials (bAPs) produced more reliable elevations in cytosolic Ca2+ concentration at distal dendritic locations in D2 MSNs than at similar locations in D1 receptor-expressing striatonigral MSNs (D1 MSNs). In both cell types, the dendritic Ca2+ entry elicited by bAPs was enhanced by pharmacological blockade of Kv4, but not Kv1 K+ channels. Local application of DA depressed dendritic bAP-evoked Ca2+ transients, whereas application of ACh increased these Ca2+ transients in D2 MSNs, but not in D1 MSNs. After DA depletion, bAP-evoked Ca2+ transients were enhanced in distal dendrites and spines in D2 MSNs. Together, these results suggest that normally D2 MSN dendrites are more excitable than those of D1 MSNs and that DA depletion exaggerates this asymmetry, potentially contributing to adaptations in PD models.
- medium spiny neuron
- glutamatergic synapse
- Parkinson's disease
- potassium channels
The principal neuronal cell type in the striatum is the medium spiny neuron (MSN). MSNs can be divided into two approximately equal groups based on axonal projections, peptide expression, and expression of dopamine (DA) receptors (Albin et al., 1989; Gerfen et al., 1998). MSNs that preferentially project axons to the substantia nigra express D1 DA receptors whereas those that preferentially project to the external segment of the globus pallidus express D2 DA receptors. Each of these MSN populations can be reliably sampled in bacterial artificial chromosome (BAC) transgenic mice in which green fluorescent protein (GFP) is expressed under control of D1 receptor or D2 receptor promoter regions (Gong et al., 2003; Day et al., 2006).
Work with these mice shows that induction of a state mimicking Parkinson's disease (PD) leads to a rapid and selective loss of spines and glutamatergic synapses in D2 receptor-expressing striatopallidal MSNs, but not in D1 receptor-expressing striatonigral MSNs (Day et al., 2006). Although dendritic Ca2+ entry through depolarization-activated Cav1.3 Ca2+ channels is necessary for the loss of spines and synapses, it is not clear why DA depletion should increase the activity of these channels. What is known is that the loss of inhibitory D2 receptor signaling in PD models selectively increases spike generation in striatopallidal MSNs (Mallet et al., 2006). In addition, D2 receptors also negatively couple to voltage dependent Cav1.3 Ca2+ channels in striatopallidal MSNs (Olson et al., 2005), suggesting that the combination of these effects might lead to increased dendritic Ca2+ entry in PD models. However, it is far from clear whether spikes initiated in the axon initial segment back-propagate any significant distance into the dendritic trees of MSNs that normally reside very near the K+ equilibrium potential (approximately −80 mV), far from spike threshold (Wilson and Kawaguchi, 1996). Work using two-photon laser scanning microscopy (2PLSM) in conjunction with patch-clamp electrophysiology has shown that proximal dendrites and spines (∼40–50 μm from the soma) of MSNs are depolarized enough by somatic spikes that voltage-dependent Ca2+ channels are opened (Carter and Sabatini, 2004). However, it is not known whether backpropagation of action potentials (bAPs) occurs in more distal dendritic regions, where most of the spine and synapse loss occurs after DA depletion. Nor is it known whether bAP invasion of the dendrites differs in D1 and D2 MSNs.
To answer these questions, 2PLSM and patch-clamp electrophysiology were used to study D1 and D2 MSNs in brain slices from BAC transgenic mice. Neurons were loaded with a Ca2+ sensitive dye to provide a direct measure of dendritic Ca2+ influx and an indirect measure of dendritic membrane potential (Carter and Sabatini, 2004). Our study suggests that normally, bAPs invade more distal dendritic regions of D2 MSNs than D1 MSNs. This invasion is controlled not only by voltage-dependent Na+ channels but also by Kv4 K+ channels. More importantly from the standpoint of understanding the adaptations in PD models, DA and acetylcholine (ACh) potently modulate the bAP-evoked dendritic Ca2+ transient in D2 MSNs, but not in D1 MSNs, providing a mechanism by which DA depletion could enhance the electrical coupling between somatic and dendritic regions and trigger spine loss. Simulations suggest that spine loss itself further increases dendritic excitability, creating the potential for progressive loss of dendritic synapses in D2 MSNs and their functional linkage with cortical structures.
Materials and Methods
Brain slice preparation.
Parasagittal brain slices (275 μm) were obtained from 17- to 25-d-old BAC D1 or BAC D2 transgenic mice following procedures approved by the Northwestern University Animal Care and Use Committee and guidelines of the National Institutes of Health. The mice were anesthetized with isoflurane (Baxter) and decapitated. Brains were rapidly removed and sectioned in oxygenated, ice-cold, artificial CSF (ACSF) using a Leica VT1000S vibratome (Leica Microsystems). The ACSF contained the following (in mm): 124 NaCl, 3 KCl, 2 CaCl2, 1 MgCl2, 26 NaHCO3, 1 NaH2PO4, and 10 d-(+)-glucose. Unless otherwise noted, all chemicals and reagents were obtained from Sigma/RBI. The slices were transferred to a holding chamber where they were incubated in ACSF at 35°C for 1 h, after which they were stored at room temperature until whole-cell recording experiments (1–5 h). The external ACSF solutions were bubbled with 95% O2/5% CO2 at all times to maintain oxygenation and a pH ∼7.4. The solutions were periodically checked and adjusted to maintain physiological osmolality (300 mOsm/L).
D1 (BAC D1) or D2 (BAC D2) receptor-expressing MSNs in 275-μm-thick corticostriatal slices were identified by somatic eGFP two-photon excited fluorescence using an Ultima Laser Scanning Microscope (2PLSM) system (Prairie Technologies). A DODT contrast detector system was used to provide a bright-field transmission image in registration with the fluorescent images. The green GFP signals (490–560 nm) were acquired using 810 nm excitation (Verdi/Mira laser: Coherent Laser Group). MSNs were patched using video microscopy with a Hitachi CCD camera (model KP-M2RN) and an Olympus UIS1 60×/0.9 NA water-dipping lens. Patch electrodes were made by pulling BF150–86-10 glass on a P-97 Flaming/Brown micropipette puller (Sutter Instrument). The pipette solution contained the following (in mm): 135 KMeSO4 (ICN Biomedicals), 5 KCl, 10 HEPES, 2 MgATP, 0.2 Na2GTP, 10 phosphocreatine, and 0.1 spermine, pH = 7.25–7.3 with KOH, 270 mOsm/L. In some experiments, the pipette solution contained the following (in mm): 120 CsMeSO3, 15 CsCl, 8 NaCl, 10 HEPES, 3 MgATP, 0.3 NaGTP, 10 TEA, 5 Qx-314. As measured in the bath, the pipette resistance was ∼4 MΩ. High-resistance (> 1 GΩ) seals were formed in voltage-clamp mode on somata. After patch rupture, the series resistance decreased to 10–15 MΩ. The inclusion of Alexa 568 (50 μm) allowed visualization of cell bodies, dendrites, and spines. After patch rupture, the internal solution was allowed to equilibrate for 15–20 min before imaging. Somata were voltage clamped at −80 mV and monitored. High-magnification maximum projection images of dendrite segments (45–150 μm) were acquired with 0.08 μm2 pixels with 10 μs dwell time; 10–20 images were taken with 0.5 μm focal steps. Maximum projection images of the soma and dendrites were acquired with 0.36 μm2 pixels with 10 μs pixel dwell time; ∼80 images were taken with 1 μm focal steps. Dendrites that were largely limited to a single optical plane (≤10 μm Z variance) were selected for calculating the bAP-evoked Ca2+ signal decrement over distance.
Electrophysiology and Ca2+ imaging.
Whole-cell current-clamp recordings were obtained using standard techniques. Slices were transferred to a submersion-style recording chamber mounted on an Olympus BX51-WIF upright, fixed-stage microscope. The slices where continuously perfused with ∼1.5 ml/min ACSF at room temperature. Electrophysiological recordings were obtained with a Multiclamp 700B amplifier (Axon Instruments) and then digitized with the scanning computer (PCI MIO-16E-4, National Instruments). The stimulation, display, and analysis software was a custom-written shareware package, WinFluor (John Dempster, Strathclyde University, Glasgow, Scotland, UK). WinFluor automated and synchronized the two-photon excited fluorescence with the electrophysiological stimulation. Single bAPs were generated by injecting current pulses (2 nA, 2 ms) at 5 s intervals. In some cases, a 1 s train of current pulses (1 nA, 2 ms) was delivered at 10 Hz. Rheobase was determined using a 1 s current step. Drugs were either bath applied by dissolving them in the external ACSF or focally applied using pressure ejection through a micropipette filled with both the drug and Alexa 568 to aid the placement of the puffer pipette and to visualize the “puff.” These compounds were dissolved in HEPES-buffers ACSF (pH = 7.3–4) and the pipette positioned 10–20 μm from the imaged dendrite/spine. α-Dendrotoxin (Alomone Labs) was dissolved in ACSF containing 0.1% BSA.
For 2PLSM imaging of Ca2+ transients, the neurons were filled via the patch pipette with the Ca2+ indicator Fluo-4 (200 μm). At this concentration, Fluo-4 gave a reliable measure of alterations in dendritic Ca2+ concentration in response to single somatic spikes, as well as short bursts of spikes. To verify that the dye was not saturated, K+ channel blockers (Cs+, 4-AP) were used to increase the dendritic depolarization; in both cases, the spike-induced fluorescence increased 3- to 10-fold, confirming that the dye was not saturated. Green fluorescent line-scan signals were acquired (as described above) at 6 ms per line and 512 pixels per line with 0.08 μm pixels and 10 μs pixel dwell time. The laser-scanned images were acquired with 810 nm light pulsed at 90 MHz (∼250 fs pulse duration). Power attenuation was achieved with two Pockels cells electro-optic modulators (models 350–80 and 350–50, Con Optics). The two cells were aligned in series to provide an enhanced modulation range for fine control of the excitation dose (0.1% steps over four decades). The line scan was started 200 ms before the stimulation protocol and continued 4 s after the stimulation to obtain the background fluorescence and to record the decay of the optical signal after stimulation. To reduce photodamage and photobleaching, the laser was fully attenuated using the second Pockels cell at all times during the scan except for the 500 ms period directly flanking the bAP.
Cell culture and immunofluorescence.
Corticostriatal cocultures were prepared to promote normal dendritic development and spine growth in medium spiny neurons (Segal et al., 2003). Striatal tissue was isolated from 1- to 2-d-old BAC D2 mice. Cortices were dissected from 18- to 19-d-old C57BL/6 mouse embryos. Tissues were digested with papain (Worthington Biochemical Corporation) and dissociated. Striatal and cortical cells were mixed at a ratio of 3:1 and plated at a density of 1 × 105/cm2 on 12 mm coverslips coated with polyethylenimine (Sigma). Coverslips were placed in 24-well plates with Neurobasal A medium (Invitrogen) supplemented with 0.5 mm glutamine (Invitrogen), 1×B27 (Invitrogen), 50 mg/L penicillin/streptomycin (Invitrogen), 50 ng/ml BDNF (Sigma), and 30 ng/ml GDNF (Sigma). After initial plating, one-fourth of the medium was exchanged with fresh medium without BDNF and GDNF every 3–4 d.
Twenty-one days after plating, cultures were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS, pH 7.4) buffer for 20 min at room temperature. Fixed cells were incubated in blocking buffer containing 0.2% Triton X-100, 1% BSA, 5% normal goat serum (Jackson ImmunoResearch Laboratories), and 0.01% sodium azide in PBS for 1 h at room temperature. Coverslips were then exposed to rabbit anti-GFP antibody (1:3000, Abcam) and mouse anti-Kv4.2 monoclonal antibody (1:500, NeuroMab) in blocking buffer overnight at 4°C. After a brief wash in PBS, cells were incubated with Alexa 488-conjugated goat anti-rabbit antibody (1:1000, Invitrogen) and Alexa 555-conjugated goat anti-mouse antibody (1:1000, Invitrogen) for 1 h at room temperature. After rinsing in PBS for 30 min, coverslips were mounted with Prolong Gold anti-fade reagent (Invitrogen). Image acquisition was performed with a LSM 510 META Laser Scanning Microscope (Zeiss).
Medium spiny neurons were modeled with NEURON, version 6.0 (Hines and Carnevale, 1997, 2001). The model neuron was constructed of a cylindrical soma (L = 12 μm, D = 12 μm), six dendrites, an axon initial segment (AIS) (L = 10 μm, D = 1.4 μm), and a cylindrical axon consisting a sequence of 5 sections (L = 100 μm, D = 1 μm) separated by 4 nodes (L = 1 μm, D = 1 μm). Each dendrite consisted of 1 primary section (L = 10 μm, D = 2.25 μm), 2 secondary sections (L = 20 μm, D = 2.5 μm), and 4 tertiary sections (L = 180 μm, D = 0.75 μm). Axial resistivity was 200 Ω · cm. Membrane capacitance was 1 μF/cm2. The cell model incorporated biophysically accurate ion channel models describing Na, BK, SK, Kv1, Kv2, Kv4, Kv7 (KCNQ), and Kir2 channels, as well as a Ca2+ buffering system, that were constrained by experimental data (Baranauskas et al., 1999; Tkatch et al., 2000; Baranauskas et al., 2003; Chan et al., 2004; Shen et al., 2005) or acquired from NEURON database mod files from previous simulations (Migliore et al., 1995; Wang et al., 2002; Khaliq et al., 2003) and incorporated into the appropriate compartments. All simulations were done at 23°C and with an ENa of 50 mV and EK of −90 mV. NEURON mod files providing a complete description of the model are available upon request and will be posted on the NeuronDB web site (http://senselab.med.yale.edu/neuronDB).
All data points represent a spine, dendrite, or spine-dendrite pair from individual cells. Data were visualized and analyzed with custom image-processing shareware software (PicViewer and WinFluor, John Dempster). IGOR Pro (WaveMetrics) was used for data smoothing and statistics. The mean fluorescence as a function of time [F(t)] was the spatial average of 5 adjacent pixels. The basal fluorescence, Fo, was average of the first 30 time points in a line scan. The percentage change in Ca2+ signal (ΔF/Fo) was defined as the maximum fluorescence change normalized by the basal fluorescence. The acquired fluorescence data were then imported to IGOR, where some traces were smoothed with a binomial Gaussian filter. The statistical significance of small, unmatched samples was determined using a nonparametric Kruskal–Wallis one-way ANOVA. Data are presented in a nonparametric format as box plots with the median shown as a central bar; upper and lower edges of the box split the upper and lower halves of the data in half again (interquartile range); the whiskers span the distribution of data points, except for outliers.
bAP-evoked Ca2+ transients decrement less in D2 than in D1 MSN dendrites
Striatopallidal (D2) or striatonigral (D1) MSNs were visually identified using 2PLSM excitation of eGFP as previously described (Day et al., 2006). Somatic whole-cell current recordings were made with electrodes filled with Alexa 568 (50 μm) and Fluo-4 (200 μm). The Alexa 568 enabled detailed visualization of distal dendrites and spines, whereas the Ca2+-sensitive indicator Fluo-4 reported Ca2+ transients induced in these regions by the somatically generated bAPs. After eGFP phenotyping and patching, the internal solution was allowed to equilibrate for 20 min. 2PLSM line scanning was then performed between 45 and 130 μm from the soma (Fig. 1A,B). Estimates of the Ca2+ transient were generated by eliciting six bAPs (equally spaced at 5 s intervals), and then averaging the responses. Spacing in this way allowed ample time for the Ca2+ to return to basal levels and remain there for several seconds before the onset on the next bAP in the series. The bAPs were evoked by injecting a 2 ms, 2 nA current pulse into the soma (Fig. 1A, right panel, I and V traces). To control for photodamage, dendritic processes were only illuminated during a 0.5 s window that bracketed the initiation of the bAP. Measurements were taken concurrently from a spine and the parent dendrite close to the base of the spine. In all cases, if a Ca2+ transient was detected in the spine, it was also detected in the dendrite. The maximum amplitude of the bAP-evoked Ca2+ transient was determined by calculating ΔF/Fo for each transient (image panels), averaging the results (black traces), and then fitting the decay phase of data with a single exponential (gray lines) to extrapolate back to the peak of the transient. The key finding in these initial experiments was that in D1 MSNs, single bAPs frequently failed to evoke a detectable Ca2+ transient at dendritic sites >60 μm from the soma (Fig. 1A), whereas in D2 MSNs, dendritic Ca2+ transients were readily detected at this distance and beyond (Fig. 1B).
To better characterize the disparity in the bAP-evoked Ca2+ transients between the two populations of MSNs, bAP-evoked Ca2+ transients from each cell type were scanned at varying distances from the soma (Fig. 1C,D). Here, the amplitudes of bAP-evoked Ca2+ transient from each scan point in each cell type were normalized to the most proximal location scanned and then plotted as a function of distance from the soma (n = 6 each). These experiments confirmed differences in somato/dendritic excitability between MSNs, with the D2 MSNs showing less attenuation of bAP-evoked Ca2+ transients in distal spines and dendrites than D1 MSNs. To test the possibility that the loss in bAP response was attributable to declining dendritic Ca2+ channel density, D1 MSNs were loaded with Cs+ (to improve voltage control of distal dendrites) and the somatic membrane briefly stepped to a depolarized potential. In this situation, there was no detectable attenuation of the Ca2+ transient with distance from the soma (Fig. 1D), arguing that the loss of the bAP-evoked Ca2+ transient was not caused by diminished Ca2+ channel density. Further evidence that this phenomenon did not simply reflect diminished Ca2+ channel density in distal dendrites was that strong depolarization (1 s) and trains of APs (10× 10 Hz) consistently evoked Ca2+ transients in distal process of all MSNs tested (supplemental Fig. S1, available at www.jneurosci.org as supplemental material).
Although single bAPs were not propagated efficiently into the distal dendrites of D1 MSNs, bursts of somatic action potentials were able to evoke Ca2+ transients in more distant dendritic regions. Three spike bursts (50 Hz) delivered at a theta frequency reliably evoked shaft and spine Ca2+ transients in both D1 and D2 MSN dendrites 100–120 μm from the soma (Fig. 1E). The Ca2+ signals evoked by successive bursts summed in a sublinear manner (Fig. 1E,F). This sublinearity was more pronounced in D1 MSNs than in D2 MSNs (Fig. 1F). Moreover, consistent with the response to single bAPs, the relative elevation in Ca2+ evoked by somatically generated theta bursts were smaller in amplitude and area in D1 MSNs (Fig. 1F).
Action potentials are actively propagated into the proximal dendrites of D2 MSNs
In the dendrites of cortical pyramidal neurons, voltage-dependent Na+ channels play an important role in actively propagating somatic spikes into dendrites (Spruston et al., 1995; Stuart et al., 1997). Blocking these channels in proximal dendrites by local application of tetrodotoxin (TTX) disrupts backpropagation of action potentials, attenuating the opening of depolarization-activated Ca2+ channels located in distal dendritic regions. To determine whether Na+ channels played a similar role in MSNs, Na+ channels were blocked by focally applying TTX (1 μm) to dendrites (Fig. 2A). Application of TTX (1 μm) to the proximal dendrite between the somatic electrode and the dendritic region being scanned consistently diminished the bAP-evoked Ca2+ transient (n = 5) (Fig. 2B) (control = black trace, TTX = red trace). Dendritic application of TTX had no effect on the somatic spike amplitude. On average, the dendritic fluorescence signal was reduced to 20% of control by application of TTX to proximal dendrites (Fig. 2B, box plot). At more distal locations (70–100 μm from the soma), TTX application had little or no effect on the bAP-associated Ca2+ transients (Fig. 2B), suggesting that voltage-dependent Na+ channels did not support bAP propagation into these more distal regions. These results suggest that bAP propagation in distal dendritic regions is decrementing in MSNs, much as in the distal dendrites of CA1 pyramidal neurons (Magee et al., 1998; Bernard and Johnston, 2003).
Kv4 K+ channels regulate dendritic excitability
The somatic excitability of MSNs is regulated by an array of K+ channels, most prominently channels with Kir2, Kv1, Kv2, Kv4, and Kv7 pore-forming subunits. Of these, Kv1 and Kv4 channels are the most likely to play a significant role in regulating dendritic bAP invasion in MSNs (Johnston et al., 2000; Tkatch et al., 2000; Shen et al., 2004). MSNs express Kv1.1, Kv1.2, and Kv1.6 channels. To test their involvement in dendritic electrogenesis, the Ca2+ transient evoked by bAPs was compared before and after bath application of α-dendrotoxin (α-DTX) (Harvey, 2001). Spines and adjacent shafts at distances from the soma where the bAP Ca2+ transient was beginning to decrement (60–120 μm) were examined. There was no change in the bAP-evoked dendritic Ca2+ transient after α-DTX (500 nm, red traces) in either D1 (Fig. 3A) or D2 MSNs (Fig. 3B) (n = 5 each), despite the fact that the somatic response to intracellular current injection was unequivocally altered by the treatment (Fig. 3C,D). In agreement with the inference that Kv1 channels were not playing a major role in regulating bAPs, 100 μm 4-aminopyridine (4-AP) also failed to alter bAP-evoked transients in dendritic shafts or spines (n = 4, data not shown).
To test for the involvement of other K+ channels, the bAP-associated dendritic Ca2+ transient was examined before and after bath application of higher concentrations of 4-AP. MSNs robustly express Kv2 and Kv4 channels (Tkatch et al., 2000; Ariano et al., 2005). As mentioned above, low concentrations of 4-AP that do not effectively block Kv4 channels (100 μm) were without effect. Elevating the 4-AP concentration to 500 μm, a concentration that should block a significant proportion of Kv2 (and Kv3) channels, also failed to significantly alter the bAP-associated Ca2+ transient (Fig. 4A) (n = 4, pink traces). However, increasing the 4-AP concentration to 1 mm (the IC50 for Kv4 channels is near 2 mm) clearly increased the dendritic Ca2+ transient. In all of the D2 MSNs tested (n = 5), 4-AP (1 mm) increased the amplitude of the bAP-evoked Ca2+ transient in proximal (∼60 μm from soma) dendritic shafts and spines (Fig. 4A,B) (black traces = control, red traces = 4-AP). Likewise, bAP-evoked Ca2+ transients were enhanced in the distal (∼120 μm from soma) dendritic shafts and spines of D1 and D2 MSNs (Fig. 4A–C). In all of the D1 MSNs tested (n = 4), 4-AP also enhanced the bAP-evoked Ca2+ transients in proximal dendritic shafts and spines, as seen in D2 MSNs (Fig. 4B). The percentage increase in the distal bAP-evoked transient by 4-AP was not calculable in D1 MSNs because the control transient was normally not detectable. Increasing the extracellular 4-AP concentration to 2 mm had an even more dramatic effect on the dendritic Ca2+ transients (traces not shown). To better evaluate the impact of Kv4 channels on the differences in the attenuation of the bAP-associated Ca2+ signal in D1 and D2 MSNs, the ratio of the distal (100–120 μm) to proximal (40–60 μm) dendritic Ca2+ transients was measured (as described above, see Fig. 1D) in the presence of 1 mm 4-AP. For the purposes of comparison, control data from Figure 1D is reproduced in Figure 4C (lightly shaded boxes). In both D1 and D2 MSNs, 4-AP eliminated the attenuation in the spine Ca2+ transient; in distal dendritic shafts the Ca2+ transient was typically larger than that seen in the proximal dendrites in the presence of 4-AP (Fig. 4C). These results suggest that Kv4 K+ channels were a major factor in the attenuation of the bAP signal.
To determine whether Kv4.2 channel proteins were properly positioned to regulate bAPs in MSNs, an immunocytochemical approach was used. Primary cell cultures were generated from the striatum of BAC D2 GFP mice and the cortex of wild-type mice (Segal et al., 2003). In these cocultures, the dendrites and spines of MSNs strongly resemble those found in vivo. However, unlike the situation in vivo, the dendrites of MSNs are planar and isolated, increasing the ability to unequivocally localize proteins with immunocytochemical approaches. Probing nonpermeabilized three-week old cocultures with an antibody directed to an extracellular epitope of the Kv4.2 subunit revealed intense labeling of soma and dendrites (Fig. 4D, red signal, left panel, top). D2 MSNs were identified by GFP fluorescence (Fig. 4D, middle panel). Overlaying the images revealed a clear expression of Kv4.2 channel protein in the shafts and spines of these MSNs (yellow, right panel, top). The labeling of distal dendritic shafts and spines was more readily seen in higher magnification images (Fig. 4D, bottom panels). Together, these results strongly suggest that Kv4.2 K+ channels are appropriately positioned to regulate dendritic excitability and bAP invasion in MSNs.
The obvious problem with the inference that Kv4 channels are important regulators of the dendritic depolarization associated with bAPs is the lack of 4-AP selectivity. Although the dose dependence of the 4-AP effect provides some measure of assurance that Kv4 channels and not Kv1 channels are involved, 4-AP is capable of blocking other K+ channels. Ba2+ also blocks Kv4 channels (Coetzee et al., 1999), but also blocks Kir2 K+ channels, which are prominent in MSNs. At present, there is no consensus on the specificity of organic toxins for Kv4 channels, undermining any pharmacological approach to the problem. As an alternative, we took a computational approach. A computational model of an MSN was created using NEURON that faithfully reproduced key features of the dendritic architecture and known intrinsic ionic mechanisms (Wilson et al., 1983) (supplemental Fig. S2, available at www.jneurosci.org as supplemental material). In our model, there was a steep Na+ channel gradient from the axon initial segment and soma into the dendrites, with tertiary dendrites being devoid of these channels, consistent with the observations described above. The density of Kv4 channels was assumed to be uniform, the most parsimonious assumption given the anatomical data on hand. Simulation of dendritic bAP invasion with this model suggest that the absence of Na+ channels in tertiary dendritic branches led to an attenuation of the dendritic depolarization produced by bAPs, with the peak of the depolarization near −45 mV in the middle portions of the tertiary branches (supplemental Fig. S2, available at www.jneurosci.org as supplemental material). This depolarization was sufficient to activate relatively low voltage-activated Cav1.3 Ca2+ channels placed in the dendrites (supplemental Fig. S2, available at www.jneurosci.org as supplemental material), as previous work has shown them to be localized in these MSN regions (Olson et al., 2005). More importantly, downregulation of Kv4 channels alone enhanced the bAP associated voltage change in distal dendrites and, in so doing, increased the opening of voltage-dependent Ca2+ channels (supplemental Fig. S2, available at www.jneurosci.org as supplemental material). The augmented Ca2+ channel opening led to a significant elevation in dendritic Ca2+ concentration, much like that seen experimentally after 4-AP application (supplemental Fig. S2, available at www.jneurosci.org as supplemental material). These simulations show that downregulation of Kv4 channel opening was sufficient to explain the elevation in dendritic Ca2+ concentration after 4-AP application, but a definitive resolution of this issue requires the development of better tools.
DA and muscarinic receptors differentially modulate dendritic Ca2+ transients
In the healthy striatum, DA and acetylcholine (ACh) (released from giant interneurons) are key modulators of synaptic integration and plasticity (Surmeier et al., 2007). Adaptations in MSN connectivity and dendritic architecture in PD models are dependent on the loss of DA and the consequent rise in ACh signaling through broadly expressed M1 muscarinic receptors (Day et al., 2006; Shen et al., 2007). How these changes in neuromodulator levels translate into changes in dendritic Ca2+ is incompletely understood. In MSN perisomatic membranes, D2 and D1 receptors have been found to couple to voltage-dependent Ca2+ channels (Surmeier et al., 1995; Hernández-López et al., 1997, 2000; Olson et al., 2005), but it is not clear that the same coupling is present in dendrites. To move toward a better grasp of these mechanisms, the ability of DA to modulate bAP-evoked Ca2+ transients was tested. In the presence of ionotropic and metabotropic glutamate and GABA receptor antagonists, local puffer application of DA (100 μm) to the dendrites of D2 MSNs reduced the bAP-associated Ca2+ transient (n = 3) (Fig. 5A, top traces). This effect was mimicked by the D2 receptor agonist quinpirole (10 μm, n = 5) (Fig. 5A, bottom traces, 5B). In contrast, DA application had no detectable effect on bAP-associated Ca2+ transients in D1 MSNs (n = 5) (Fig. 5B). This result argues that the negative coupling of D2 receptors to Ca2+ channels in perisomatic membrane (Hernández-López et al., 2000) extends into the dendritic shafts and spines of D2 MSNs.
In the perisomatic membrane of MSNs, muscarinic receptors also negatively couple to Ca2+ channels. There, M1 muscarinic receptor signaling in both D2 and D1 MSNs diminishes the opening of Cav1 channels, whereas M4 receptor signaling decreases the opening of Cav2 channels preferentially in D1 MSNs (Howe and Surmeier, 1995; Yan et al., 2001). In the same mixture of glutamate and GABA receptor antagonists, focal application of muscarine (20 μm) to the dendrites of D2 MSNs (60–100 μm from the soma) significantly increased (not decreased) the bAP-associated Ca2+ transient (n = 5) (Fig. 5C,D). In contrast, puffing muscarine on the proximal portion of the D1 MSNs dendritic tree (45–60 μm from the soma) had no effect on bAP associated Ca2+ transients in spines (n = 5) (Fig. 5D). In pyramidal neurons, M1 muscarinic receptor signaling potently downregulates Kv4 K+ channel opening in response to membrane depolarization and increases bAP amplitude in distal dendritic regions (Hoffman and Johnston, 1998; Yuan et al., 2002). If M1 muscarinic receptor signaling was doing something similar in MSNs to enhance Ca2+ channel opening, then blocking Kv4 channels should occlude the effects of M1 receptor stimulation. To test this hypothesis, 4-AP (1 mm) was bath applied and then muscarine was puffed on the dendrites of D2 MSNs. As predicted, in this situation muscarine had little or no effect on the bAP-evoked change in dendritic fluorescence (n = 5) (Fig. 5C,D), suggesting that Kv4 channels are dendritic targets of muscarinic receptor signaling.
DA depletion enhances bAP invasion in D2 MSNs
The experiments described thus far show that DA-receptor signaling diminishes bAP-evoked Ca2+ transients, whereas ACh-receptor signaling increases bAP-evoked Ca2+ transients in D2 MSNs. In PD models, striatal DA levels fall and ACh signaling rise, suggesting that there should be a profound increase in dendritic excitability in D2 MSNs, at least acutely. What happens with sustained DA depletion is less clear. To pursue this question, BAC D2 mice were DA depleted for 5 d using reserpine (LaHoste et al., 1993; Kachroo et al., 2005). Previous work by our group has shown that this produces a profound loss of spines in D2 MSNs that is mimicked by 6-hydroxydopamine lesions of the nigrostriatal dopaminergic system (Day et al., 2006). This spine pruning requires the opening of depolarization-activated Cav1.3 Ca2+ channels and is attenuated by genetically deleting M1 muscarinic receptors (Day et al., 2006; Shen et al., 2007). In brain slices from DA-depleted mice, the bAP-evoked Ca2+ transient was mapped in the dendrites of D2 MSNs. As described above, the amplitude of the fluorescence change (ΔF/F0) at distal dendritic sites was normalized by the proximal fluorescence signal. In D2 MSNs from DA-depleted mice, the relative amplitude of bAP-evoked Ca2+ transient in dendritic shafts and spines fell less steeply with distance from the soma than in untreated neurons (n = 4) (Fig. 6A). At distal dendritic locations (100 and 150 μm from the soma), DA depletion significantly increased the relative amplitude of the Ca2+ transient evoked by a single bAP (Fig. 6B) (n = 4 each). In fact, in all of the neurons examined after DA depletion, bAP-associated Ca2+ transients were detectable as far out on the dendrites as we were capable of imaging (∼150 μm from the soma).
Based on the work described above in the normosensitive striatum, the simplest explanation of this change is that it reflects the loss of D2 receptor and the gain of M1 receptor activity after reserpine treatment. As a first step toward testing this hypothesis, the amplitudes of bAP-evoked Ca2+ transients in untreated BAC D2 MSNs were recorded before and after bath application of the D2 antagonist sulpiride (10 μm) to determine whether ambient DA release in the brain slice contributed to the profile of dendritic excitability measured. There was not any significant change in the bAP-evoked Ca2+ transient in distal dendrites after D2 blockade, indicating that ambient DA release was not a factor (n = 5, data not shown). To test for the possibility that DA depletion elevated M1 muscarinic receptor activity, the muscarinic antagonist scopolamine (20 μm) was bath applied to slices from reserpinized mice; scopolamine significantly reduced the bAP-evoked Ca2+ transient in D2 MSNs (n = 5) (Fig. 6C).
Another potential factor in the enhanced dendritic excitability of D2 MSNs after DA depletion is the loss of spines and dendritic surface area itself. This loss should diminish the capacitative load of the dendrites and improve bAP invasion into distal regions. Although consistent with theoretical and experimental examination of other neurons (Wilson, 1992), this hypothesis was tested in an anatomically representative model of an MSN. NEURON simulations were conducted in which the surface area of spiny dendrites was decreased and the effects on the bAP examined. These simulations corroborated the inference that spine loss enhances dendritic bAP invasion, showing enhanced bAP propagation, enhanced opening of voltage-dependent Ca2+ channels and an elevation in bAP-evoked change in intracellular Ca2+ concentration at distal dendritic locations (supplemental Fig. S3, available at www.jneurosci.org as supplemental material).
Striatal MSNs are the principal neurons of the striatal circuitry that controls a wide array of psychomotor behaviors. Yet, relatively little is known about how intrinsic dendritic mechanisms govern the integration of synaptic signals. A major obstacle to gaining a better understanding of these regions is their small size and largely nonplanar organization. Optical approaches, particularly 2PLSM, offer a powerful strategy for probing dendritic function, particularly when used in conjunction with somatic patch-clamp recordings. This approach has recently been used to study proximal MSN dendrites (Carter and Sabatini, 2004; Carter et al., 2007). Our results extend these observations to more distal dendritic regions of MSNs and to differences between the two major subtypes of striatal MSNs. Our studies support four basic conclusions. First, bAPs are actively propagated in the proximal dendritic trees of MSNs, but appear to be passively propagated into more distal dendritic regions; moreover, this propagation is more robust in D2 MSNs than in D1 MSNs. Second, the propagation of potential changes produced by bAPs are actively shaped by dendritic K+ channels, most likely Kv4 channels, as has been described in CA1 pyramidal neurons (Bernard and Johnston, 2003). Third, the dendritic Ca2+ signal associated with a single bAP is modulated by focal application of both DA and ACh receptor agonists in D2 MSNs, but this bAP-associated signal is not reliably modulated in D1 MSNs. Last, DA depletion increases the dendritic Ca2+ signal associated with bAPs in D2 MSNs, adding an important new insight into the mechanisms underlying striatal adaptations in PD.
Dendrites of D2 MSNs are more excitable than those of D1 MSNs
Individual action potentials generated at the soma produced reliable Ca2+ transients in proximal (∼30–50 μm from the soma) dendritic shafts and spines of both D2 and D1 MSNs. These bAP-evoked Ca2+ transients were also reliably detected in more distal (∼100 μm from to soma) dendrites and spines of the D2 MSNs. The relative amplitude of this fluorescence signal fell with distance beyond ∼50 μm from the soma, presumably because the amplitude of the bAP-associated potential change also fell with eccentricity from the soma (Stuart and Sakmann, 1994; Spruston et al., 1995). The rate at which this signal fell with distance was significantly greater in D1 MSNs than in D2 MSNs. The attenuation of the dendritic Ca2+ transient with distance from the soma was not attributable to a parallel fall-off in the density of Ca2+ channels, as improving distal voltage control by filling cells with Cs+ eliminated any obvious attenuation of the dendritic Ca2+ signal with distance from the soma. Rather, the attenuation was more likely to be caused by decrementing propagation of the bAP-evoked potential into distal dendrites. This inference is drawn from the observation that application of the Na+ channel toxin TTX to the proximal dendrites virtually eliminated more distal bAP-evoked elevations in Ca2+ dependent fluorescence (indicating active propagation of the bAP through the proximal dendrites), whereas application of TTX to distal tertiary dendritic locations (∼60 μm from the soma) had virtually no effect on bAP-evoked fluorescence changes. Computer simulations using a model that captured key features of the MSN geometry and channel expression, confirmed that in tertiary dendrites lacking Na+ channels, bAPs declined in amplitude as they traveled away from the soma. But these simulations also suggested that, at least within the initial portion of the tertiary dendrites, the amplitude of the bAP was still sufficient to activate relatively low-threshold Cav1.3 or Cav3 Ca2+ channels (Carter and Sabatini, 2004).
Why there appears to be a greater attenuation of bAP propagation into the dendrites of D1 MSNs is entirely unclear. Studies in other neurons have shown that dendritic geometry is an important factor governing bAP propagation (Vetter et al., 2001; Schaefer et al., 2003). However, a recent study by our group (Gertler et al., 2008) failed to find any significant differences in the branching structure of D1 and D2 MSN dendrites, although D1 MSNs had more primary dendrites. In agreement with this anatomical similarity, the electrotonic length of D1 and D2 MSN dendrites were indistinguishable. Another factor governing bAP propagation is the dendritic distribution of ion channels. Voltage-dependent Na+ channels support bAPs, helping to maintain the amplitude of bAPs as they invade dendrites (Stuart and Sakmann, 1994). Voltage-dependent K+ channels, on the other hand, oppose bAP propagation (Hoffman et al., 1997). Kir2 K+ channels are robustly expressed in MSN dendrites (Prüss et al., 2005; Shen et al., 2007); however, inwardly rectifying Kir2 channels rapidly block at potentials above the K+ equilibrium potential, making them poor regulators of bAP propagation. In other neurons, depolarization-activated Kv4 channels have been shown to be potent bAP regulators (Hoffman et al., 1997). Our work revealed that the dendrites of MSNs also are invested with Kv4 channels, in agreement with previous scRT-PCR studies showing Kv4.1–3 mRNA expression in MSNs (Tkatch et al., 2000). Moreover, block of Kv4, but not Kv1, channels enhanced bAP-evoked dendritic Ca2+ signals. Simulations of bAP propagation suggested that reducing Kv4 density by half could readily account for the change in dendritic bAP-evoked Ca2+ signals. More importantly, partially blocking Kv4 K+ channels with 4-AP eliminated the attenuation in the Ca2+ transient in distal dendritic regions in both D1 and D2 MSNs. However, it is not clear that the differences in the dendritic excitability of D1 and D2 MSNs is directly dependent on Kv4 channel density or function. Alternative approaches (Kim et al., 2007) will be necessary to unequivocally answer this question.
DA suppresses, whereas ACh enhances, dendritic excitability in the D2, but not D1, MSNs
Asymmetries in the neuromodulatory effects of DA on striatopallidal and striatonigral MSNs have long been inferred from their differential expression of D1 and D2 receptors. D1 receptor stimulation generally enhances the response to excitatory inputs, whereas D2 receptor stimulation attenuates responses to excitatory stimulation (Levine et al., 1996; Cepeda et al., 1998). Our results point to another example, showing that local application of DA diminishes bAP-evoked Ca2+ signals in the dendrites of striatopallidal D2 MSNs, but has no detectable effect on the same response in striatonigral D1 MSNs. The D2 receptor-mediated response in D2 MSNs is consistent with their negative coupling to the Cav1 Ca2+ channels likely to underlie the bAP-evoked response (Hernández-López et al., 2000; Carter and Sabatini, 2004; Olson et al., 2005). Although ambient D2 receptor stimulation was not a factor underlying the asymmetry between MSNs (D2 receptor antagonism had no effect on basal excitability in the slice), in vivo ongoing D2 receptor activity should reduce the differences in dendritic Ca2+ signaling between MSNs.
The absence of a dendritic response to DA application in D1 MSNs is somewhat surprising. D1 receptor protein is clearly present in the dendrites of MSNs (Hersch et al., 1995). Moreover, D1 receptor stimulation modulates ion channels that regulate dendritic Ca2+ transients. D1 receptor stimulation promotes the slow inactivation of Na+ channels in MSNs (Calabresi et al., 1987; Carr et al., 2003); however, because slow inactivation only occurs at depolarized potentials, our experimental paradigm was not suited to bringing out this modulation. D1 receptor signaling also downregulates Cav2 and upregulates Cav1 Ca2+ channel opening in acutely isolated MSNs (Surmeier et al., 1995; Olson et al., 2005). Previous work also has shown that the D1 receptor-mediated enhancement of NMDA responses is dependent on Cav1 channels (Cepeda et al., 1998). The failure to detect a clear effect of D1 agonists on bAP-evoked Ca2+ transients could be attributable to several experimental factors (e.g., disruption of intracellular signaling because of dialysis), but the most likely explanation is that this modulation is not effectively assayed by a single bAP.
In contrast to DA, it has generally been thought that ACh affects both classes of MSN similarly. All MSNs robustly express M1 muscarinic receptors (Yan et al., 2001). The other muscarinic receptor expressed by MSNs, the M4 receptor, is present in both classes, albeit at significantly higher levels in striatonigral D1 MSNs (Bernard et al., 1992; Yan et al., 2001). Electrophysiological studies of muscarinic effects in randomly sampled MSNs have not found a pronounced heterogeneity in responses (Akins et al., 1990; Hersch et al., 1994; Galarraga et al., 1999). However, more recent work with BAC transgenic mice has found a much stronger M1 receptor-mediated modulation of dendritic Kir2 channels in D2 MSNs than in the D1 MSNs, a difference attributable to the susceptibility of targeted channels, not upstream signaling (Shen et al., 2007). Here, bAP-evoked dendritic Ca2+ transients were enhanced in D2 MSNs by local application of a muscarinic agonist, but not in D1 MSNs. This modulation was occluded by 4-AP, suggesting that the modulation was mediated by M1 receptor coupling to Kv4 channels, as found in pyramidal neurons (Hoffman and Johnston, 1998; Yuan et al., 2002). Why striatonigral D1 MSNs should be unresponsive despite their expression of functional M1 receptors and the dendritic localization of Kv4 channels is not clear.
Dendritic excitability of striatopallidal MSNs is enhanced in PD models
In mouse models of PD, D2 receptor-expressing striatopallidal MSNs, but not D1 receptor-expressing striatonigral MSNs, undergo a dramatic pruning of dendritic spines and synapses (Day et al., 2006). The reduction in spines requires activation of L-type Cav1.3 Ca2+ channels that are dendritically positioned through a scaffolding interaction with Shank (Zhang et al., 2005). How DA depletion increases the opening of dendritic Cav1.3 channels is uncertain. In acutely isolated MSNs, D2 receptor signaling decreases the open probability of Cav1.3 channels (Hernández-López et al., 2000; Olson et al., 2005). DA depletion would remove this inhibitory modulation if it existed in dendrites. Our results are consistent with this possibility, showing that dendritic D2 receptors decrease bAP-evoked Ca2+ transients.
Two additional factors are likely to contribute to increased Cav1.3 channel opening after DA depletion. First, DA depletion elevates cholinergic signaling in the striatum (Kopin, 1993). This should downregulate dendritic Kir2 and Kv4 K+ channels in D2 MSNs, increasing the dendritic depolarization produced by glutamatergic synapses and increasing bAP invasion into distal dendrites. The impact of elevated cholinergic signaling on bAP-evoked Ca2+ transients after DA depletion was evident in our experiments. The importance of these M1 receptor-mediated effects is underscored by attenuation of dendritic remodeling in M1 receptor knock-out mice after DA depletion (Shen et al., 2007). In vivo, where M1 receptor tone is undoubtedly higher than in the slice, D2 MSN dendrites could be even more excitable after DA depletion. Second, the loss of spines and dendritic surface area should in and of itself elevate dendritic excitability by decreasing capacitive loading. Our simulations were consistent with this inference. This creates a potentially pathological positive feedback that could induce a progressive loss of spines and synapses in PD. This combination of mechanisms provides a framework within which the synaptic pruning seen in D2 MSNs of PD models can be understood and establishes potential therapeutic targets for PD patients.
This work was supported by National Institutes of Health (NIH) Grant NS34696 (D.J.S.), the Picower Foundation (D.J.S.), and NIH Grant MH76164 (M.D.). We thank Drs. John Dempster, Philip Hockberger, Qing Ruan, Joshua Held, Karen Saporito, and Sasha Ulrich for excellent technical assistance, and Dr. Nelson Spruston for careful reading of this manuscript.
- Correspondence should be addressed to D. James Surmeier, Department of Physiology, Feinberg School of Medicine, Northwestern University, 303 East Chicago Avenue, Chicago, IL 60611.